[Alexander-Mitrofanov]

Closes #4. Targets issue-1-python3-bugfixes.

Prepares CopraRNA 3.0.0 as a Python-only release. Genome availability no longer depends on periodically generated tables: the workflow resolves current complete RefSeq assemblies on demand, caches exact chromosome/plasmid versions, and rejects duplicate assemblies. coprarna genomes supports inspection, prefetching and explicit refresh.

• Remove the original scripts, catalogs, migration sidecars, interpreter environments, script aliases and defect-emulation paths. Keep independently captured scientific expectations and provenance unchanged, protected by checksums.
• Normalize sequence, plasmid and assembly input IDs to primary-chromosome IDs in a separate working FASTA. Preserve original inputs, versioned genome provenance and compact native DomClust identifiers.
• Refresh all 17 reference GenBank records from NCBI, store them as versioned gzip files, and record download URLs, dates, sequence lengths, CDS counts and hashes. Include a staged Python refresh tool.
• Replace README.md with the complete version 3 guide; remove README.2.md and update ChangeLog, architecture policy, packaging, CI and version reporting.

Validation: 175 tests passed, including real IntaRNAsTar/BLAST+/DomClust workflows, previously unseen NZ_ and assembly inputs, all 51 reference-region extractions, cache/pagination/failure contracts, exact retained numerical fixtures and Chromium export. The final wheel and sdist build successfully and contain no retired source/catalog files. An installed wheel completed a native prediction outside the source checkout. Live NCBI lookups and all 17 genome downloads were also verified. Numerical dependency pins are unchanged.

Metadata and annotations stay cached until explicitly refreshed; new data/settings should use a fresh job directory. Supported online resolutions require current complete RefSeq assemblies. Original source history remains available in Git.